DOCSLIB.ORG

BIPN140 Lecture 11: Synaptic Plasticity (I)

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
BIPN140 Lecture 11: Synaptic Plasticity (I) BIPN140 Lecture 11: Synaptic Plasticity (I) 1. Short-Term Synaptic Plasticity 2. Long-Term Synaptic Plasticity 3. Molecular Mechanisms of Long-Term Potentiation Su (FA16) Synaptic Plasticity Synaptic plasticity: the connection between neurons are subject to modification (development, activity, physiological state, experience, disease, drugs, etc.). Changes in synaptic efficacy are widely believed as the basis for: (1) certain forms of learning (2) storage of memories (3) refinement of connections during development How to modify the strength/nature of connection between neurons? (1) changes in the number of synapses (2) changes in the probability of NT release (presynaptic mechanism) (3) changes in the number and/or property of NT receptors (postsynaptic mechanism) Synaptic plasticity varies in time scale from milliseconds to years. Short-term Synaptic Plasticity Most short-term synaptic plasticity (milliseconds to minutes) affect the amount the neurotransmitter released from presynaptic terminals. Key regulators: frequency of presynaptic APs, presynaptic [Ca2+] /sensitivity to Ca2+, availability of readily releasable SVs. Four major types: (1) Synaptic facilitation: ms, Ca2+ build up due to quick succession of APs (2) Synaptic depression: ms, transient depletion of readily releasable SVs. (3) Synaptic augmentation: seconds, more effective use of Ca2+. (4) Post-tetanic potentiation (PTP): seconds to minutes, better mobilization of SV pools (more phosphorylation of synapsin to mobilize SVs) Synaptic Facilitation (Fig. 8.1) AP interval A rapid increase in synaptic strength that occurs when two or more APs invade the presynaptic terminal within a few milliseconds. Facilitation by the 1st AP (conditioning pulse) lasts for tens of milliseconds. Prolonged elevation of presynaptic Ca2+ level (Ca2+ influx is faster than Ca2+ removal), resulting in Ca2+ buildup. 2+ n At NMJ, EPP amplitude = K([Ca ]o) , n=3.8 (Four Ca2+ ions cooperate to cause release of synaptic vesicles) Possible molecular mechanism: cooperativity of synaptotagmin-Ca2+ binding, partial occupancy of the Ca2+ binding site enhances further Ca2+ /phospholipid binding. PSPs rise Paired-pulse facilitation: 2+ depends on AP frequency, sharply as Ca increases. Ca2+ concentration Favorable experimental condition for synaptic facilitation: lowering external Ca2+. Why? Synaptic Depression (Fig. 8.1) 2+ Decline of neurotransmitter release during sustained [Ca ]out = 0.28 mM synaptic activity. 2+ [Ca ]out = 2.5 mM The occurrence/degree of facilitation or depression depends on extracellular Ca2+ concentration. High frequency APs Depression depends on the amount of neurotransmitter that has been released. Possible mechanism: depression is caused by progressive depletion of a pool of synaptic vesicles that are available for release. Experimental support: more depression is observed after the size of the reserve pool is reduced in synapsin knockout animals. Paired-pulse depression: also depends on AP frequency Augmentation & Potentiation (Fig. 8.1) Forms of short-term plasticity that occur over a slightly longer time scale (seconds to minutes). Elicited by repetitive synaptic activity that lead to prolonged elevation of presynaptic Ca2+. Function: increase the amount of neurotransmitter released from presynaptic terminals. Both enhance the efficacy of Ca2+ to trigger SV release. Augmentation: rises and falls over a few seconds (possibly regulated by munc-13, a priming factor, PLC dependent, increase the size of readily releasable SV pool and release probability ) Potentiation: acts over tens of seconds to minutes (via kinases that phosphorylate substrates such as synapsin to increase the size of readily releasable SV pool and release probability). Short-term Plasticity at the Neuromuscular Synapse (Fig. 8.2) During repetitive synaptic activity, different forms of short-term plasticity can interact to cause complex & dynamic changes of synaptic activity. Readily releasable SVs: if many are still available, facilitation (immediate) and augmentation (slightly later). If depleted, synaptic depression. Later on, PTP increases the size of readily releasable pool to further potentiate the synapse. Synaptic transmission can change dynamically as a consequence of the recent history of synaptic activity (context or experience). Long-term Synaptic Plasticity Long-term synaptic plasticity alter synaptic transmission over time scales of 30 minutes or longer. Arises from molecular mechanisms that vary over time (short to long): (1) post-translational modifications (e.g. phosphorylation) of existing proteins (changes in AMPA receptor trafficking) (2) changes in translations (local translation) (3) changes in gene transcription + translation (producing enduring changes in synaptic transmission). Two major forms: (1) long-term potentiation (LTP): long-lasting increase in synaptic strength (2) long-term depression (LTD): long-lasting decrease in synaptic strength Changes in synaptic efficacy that are likely the basis for: (1) certain forms of learning (2) storage of memories (3) refinement of synaptic connections during development Hippocampus (Fig. 8.6) A cortical structure in the medial portion of the temporal lobe (medial temporal lobe in primates). Critical for the consolidation of information from short-term memory to long-term memory and spatial navigation & spatial memory. Different types of neurons and collaterals (axon bundles) are organized in an orderly and relatively planar manner that is conducive to slice preparation electrophysiology. Most of the relevant circuitry is left intact in a hippocampus slice. The Tri-synaptic Circuit of the Hippocampus (Fig. 8.6) Tri-synaptic circuit: 3 axon bundles (perforant path, mossy fibers & Schaffer collaterals) + 3 main neuronal types (granule cells, CA3 & CA1 pyramidal neurons) Input from entorhinal cortex (via perforant path) => granule cell (via mossy fibers) => CA3 pyramidal cell (via Schaffer collaterals) => CA1 pyramidal cell. Excitatory glutamatergic connections. Clusters of monosynaptic connections. CA: cornu ammonis (ram’s horn) Long-term Potentiation (LTP) Extracellular recording Granule cells High frequency stimulation Perforant path Inducing LTP: a tetanus (high-frequency A few seconds of high-frequency electrical electrical stimulus, e.g. 100 Hz). stimulation can enhance synaptic transmission in the rabbit hippocampus for Recording: extracellular recording to days. measure amplitude or the slope of field potential (collective EPSPs) or intracellular patch-clamp recording to measure the amplitude of EPSPs. LTP is Input-Specific and Persistent (Fig. 8.7) Tetanus Tetanus stimulus at Schaffer collaterals, patch- clamp recordings at CA1 pyramidal neurons. If the Schaffer collaterals are stimulated only 2-3 times per minute, the size of the evoked EPSP in the CA1 neurons remain constant (pre-tetanus stimulus, baseline EPSP). Tetanus at pathway 1 induced LFP when pathway 1 In vivo recording with embedded electrode is stimulated. As a control, activation of pathway 2 (no tetanus) does not yield EPSPs of enhanced amplitude (no LTP) on the same postsynaptic neuron. LTP is input-specific. LTP can last for a long time! LTP is Associative (Fig. 8.9) S1 From Bliss & Collingridge (1993), Nature 361, 31-39. S1 is the weak input (low intensity), while S2 is (PTP) the strong one (high intensity). Tetanus in S1 does not do much to EPSPs (no LTP), but S2 S1+S2 together result in LTP in both synapses. A form of hetero-synaptic LTP. Associative learning? Classical Pavlovian conditioning. Independent stimulation Coordinated stimulation of pathway 1 & 2 of pathway 1 & 2 Weak stimulation of pathway 2 alone does not trigger LTP. However, when the same stimulus to pathway 2 is activated together with strong activation of pathway 1, both sets of synapses are strengthened. LTP involves Coincidence Detection (Fig. 8.8) Associative LTP suggests that LTP occurs when both presynaptic and postsynaptic neurons are active. That is, the presynaptic neurons fires action potentials and the postsynaptic membrane is depolarized. (in the CA1 neurons) To test, low intensity tetanus at Schaffer collateral => no LTP in CA1 neurons. Paired the low-intensity stimulus with strong depolarization of the CA1 neurons, within 100 ms of neurotransmitter release from the Schaffer collaterals => result in LTP. LTP occurs only when presynaptic action potentials and postsynaptic depolarization are tightly linked in time => a coincidence detector. LTP: Potential for Information Storage Input-specific: when LTP is induced by activation of one synapse, it does not occur in other, inactive synapses that contact the same neuron. Selective storage of information likely takes place at synapses, not neurons. State-dependent: pairing of pre- and postsynaptic events, coincidence detector. Hebbian Principle: Donald Hebb, a Canadian psychologist, in 1949 stated “when an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A ' s efficiency, as one of the cells firing B, is increased”. In other words, coordinated activity of a presynaptic terminal and postsynaptic neuron would strengthen the synaptic connection between them, as observed in LTP much later. A simplified version of Hebb’s rule: neurons that fire together wire together. LTP at synapses shows many properties for learning. Mechanism I: NMDA Receptors
Load more

Recommended publications
  • Song Exposure Affects HVC Ultrastructure in Juvenile Zebra Finches
    Song Exposure affects HVC Ultrastructure in Juvenile Zebra Finches Houda G Khaled Advisor: Dr. Sharon M.H. Gobes Department of Neuroscience Submitted in Partial Fulfillment of the Prerequisite for Honors in Biochemistry May 2016 © 2016 Houda Khaled Table of Contents ABSTRACT ........................................................................................................................................................ 2 ACKNOWLEDGEMENTS .............................................................................................................................. 3 INTRODUCTION ............................................................................................................................................. 4 I. SYNAPTIC PLASTICITY IN LEARNING AND MEMORY ................................................................................ 4 Basic principles of synapse structure ............................................................................................................. 4 Changes in synapse number, size and shape ................................................................................................ 6 II. SONG ACQUISITION AS A LEARNING PARADIGM .................................................................................. 7 The song system ................................................................................................................................................. 9 Structural synaptic plasticity in the song system .....................................................................................
    [Show full text]
  • Neuromodulators and Long-Term Synaptic Plasticity in Learning and Memory: a Steered-Glutamatergic Perspective
    brain sciences Review Neuromodulators and Long-Term Synaptic Plasticity in Learning and Memory: A Steered-Glutamatergic Perspective Amjad H. Bazzari * and H. Rheinallt Parri School of Life and Health Sciences, Aston University, Birmingham B4 7ET, UK; [email protected] * Correspondence: [email protected]; Tel.: +44-(0)1212044186 Received: 7 October 2019; Accepted: 29 October 2019; Published: 31 October 2019 Abstract: The molecular pathways underlying the induction and maintenance of long-term synaptic plasticity have been extensively investigated revealing various mechanisms by which neurons control their synaptic strength. The dynamic nature of neuronal connections combined with plasticity-mediated long-lasting structural and functional alterations provide valuable insights into neuronal encoding processes as molecular substrates of not only learning and memory but potentially other sensory, motor and behavioural functions that reflect previous experience. However, one key element receiving little attention in the study of synaptic plasticity is the role of neuromodulators, which are known to orchestrate neuronal activity on brain-wide, network and synaptic scales. We aim to review current evidence on the mechanisms by which certain modulators, namely dopamine, acetylcholine, noradrenaline and serotonin, control synaptic plasticity induction through corresponding metabotropic receptors in a pathway-specific manner. Lastly, we propose that neuromodulators control plasticity outcomes through steering glutamatergic transmission, thereby gating its induction and maintenance. Keywords: neuromodulators; synaptic plasticity; learning; memory; LTP; LTD; GPCR; astrocytes 1. Introduction A huge emphasis has been put into discovering the molecular pathways that govern synaptic plasticity induction since it was first discovered [1], which markedly improved our understanding of the functional aspects of plasticity while introducing a surprisingly tremendous complexity due to numerous mechanisms involved despite sharing common “glutamatergic” mediators [2].
    [Show full text]
  • Review Article Mechanisms of Cerebrovascular Autoregulation and Spreading Depolarization-Induced Autoregulatory Failure: a Literature Review
    Int J Clin Exp Med 2016;9(8):15058-15065 www.ijcem.com /ISSN:1940-5901/IJCEM0026645 Review Article Mechanisms of cerebrovascular autoregulation and spreading depolarization-induced autoregulatory failure: a literature review Gang Yuan1*, Bingxue Qi2*, Qi Luo1 1Department of Neurosurgery, The First Hospital of Jilin University, Changchun, China; 2Department of Endocrinology, Jilin Province People’s Hospital, Changchun, China. *Equal contributors. Received February 25, 2016; Accepted June 4, 2016; Epub August 15, 2016; Published August 30, 2016 Abstract: Cerebrovascular autoregulation maintains brain hemostasis via regulating cerebral flow when blood pres- sure fluctuation occurs. Monitoring autoregulation can be achieved by transcranial Doppler ultrasonography, the pressure reactivity index (PRx) can serve as a secondary index of vascular deterioration, and outcome and prognosis are assessed by the low-frequency PRx. Although great changes in arterial blood pressure (ABP) occur, complex neu- rogenic, myogenic, endothelial, and metabolic mechanisms are involved to maintain the flow within its narrow limits. The steady association between ABP and cerebral blood flow (CBF) reflects static cerebral autoregulation (CA). Spreading depolarization (SD) is a sustained depolarization of neurons with concomitant pronounced breakdown of ion gradients, which originates in patients with brain ischemia, hemorrhage, trauma, and migraine. It is character- ized by the propagation of an extracellular negative potential, followed by an increase in O2 and glucose consump- tion. Immediately after SD, CA is transiently impaired but is restored after 35 min. This process initiates a cascade of pathophysiological mechanisms, leading to neuronal damage and loss if consecutive events are evoked. The clini- cal application of CA in regulating CBF is to dilate the cerebral arteries as a compensatory mechanism during low blood pressure, thus protecting the brain from ischemia.
    [Show full text]
  • Interplay of Multiple Plasticities and Activity Dependent Rules: Data, Models and Possible Impact on Learning
    Interplay of multiple plasticities and activity dependent rules: data, models and possible impact on learning Gaetan Vignoud, Alexandre Mendes, Sylvie Perez, Jonathan Touboul*, Laurent Venance* (* equal contributions, email: fi[email protected]) College` de France, Center for Interdisciplinary Research in Biology, 11 Place Marcelin Berthelot, 75005 Paris, France Abstract the millisecond timing of the paired activities on either side of Synaptic plasticity, the activity-dependent evolution neuronal the synapse (Feldman, 2012). As such, plasticity in the stria- connectivity, is admitted to underpin learning and memory. A tum has been widely studied and a variety of spike-timing de- paradigmatic plasticity depending on the spike timings of cells pendent plasticity were observed in response to a large num- on both sides of the synapse (STDP), was experimentally ev- ber of presentations of a fixed pre- and post-synaptic stim- idenced through multiple repetitions of fixed pre- and post- ulation pattern. Those plasticities were attributed to a di- synaptic spike patterns. Theoretical and experimental com- verse neurotransmitters and pathways including for instance munities often implicitly admit that plasticity is gradually estab- endocannabinoids (eCB), NMDA receptors, AMPA receptors lished as stimulus patterns are presented. Here, we evaluate or voltage-sensitive calcium channels. Moreover, the vast ma- this hypothesis experimentally and theoretically in the stria- jority of the studies focused on cortical inputs, and much
    [Show full text]
  • Long-Term Potentiation and Long-Term Depression of Primary Afferent Neurotransmission in the Rat Spinal Cord
    The Journal of Neuroscience, December 1993. 13(12): 52286241 Long-term Potentiation and Long-term Depression of Primary Afferent Neurotransmission in the Rat Spinal Cord M. RandiC, M. C. Jiang, and R. Cerne Department of Veterinary Physiology and Pharmacology, Iowa State University, Ames, Iowa 50011 Synaptic transmission between dorsal root afferents and ably mediated by L-glutamate, or a related amino acid (Jahr and neurons in the superficial laminae of the spinal dorsal horn Jessell, 1985; Gerber and RandiC, 1989; Kangrga and Randic, (laminae I-III) was examined by intracellular recording in a 1990, 1991; Yoshimura and Jessell, 1990; Ceme et al., 1991). transverse slice preparation of rat spinal cord. Brief high- Neuronal excitatory amino acids (EAAs), including gluta- frequency electrical stimulation (300 pulses at 100 Hz) of mate, produce their effects through two broad categoriesof re- primary afferent fibers produced a long-term potentiation ceptors called ionotropic and metabotropic (Honor6 et al., 1988; (LTP) or a long-term depression (LTD) of fast (monosynaptic Schoepp et al., 1991; Watkins et al., 1990). The ionotropic and polysynaptic) EPSPs in a high proportion of dorsal horn NMDA, a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic neurons. Both the AMPA and the NMDA receptor-mediated acid (AMPA)/quisqualate (QA), and kainate receptors directly components of synaptic transmission at the primary afferent regulate the opening of ion channelsto Na, K+, and, in the case synapses with neurons in the dorsal horn can exhibit LTP of NMDA receptors, CaZ+as well (Mayer and Westbrook, 1987; and LTD of the synaptic responses. In normal and neonatally Ascher and Nowak, 1987).
    [Show full text]
  • Cerebral Pressure Autoregulation in Traumatic Brain Injury
    Neurosurg Focus 25 (4):E7, 2008 Cerebral pressure autoregulation in traumatic brain injury LEONARDO RANGE L -CASTIL L A , M.D.,1 JAI M E GAS C O , M.D. 1 HARING J. W. NAUTA , M.D., PH.D.,1 DAVID O. OKONK W O , M.D., PH.D.,2 AND CLUDIAA S. ROBERTSON , M.D.3 1Division of Neurosurgery, University of Texas Medical Branch, Galveston; 3Department of Neurosurgery, Baylor College of Medicine, Houston, Texas; and 2Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania An understanding of normal cerebral autoregulation and its response to pathological derangements is helpful in the diagnosis, monitoring, management, and prognosis of severe traumatic brain injury (TBI). Pressure autoregula- tion is the most common approach in testing the effects of mean arterial blood pressure on cerebral blood flow. A gold standard for measuring cerebral pressure autoregulation is not available, and the literature shows considerable disparity in methods. This fact is not surprising given that cerebral autoregulation is more a concept than a physically measurable entity. Alterations in cerebral autoregulation can vary from patient to patient and over time and are critical during the first 4–5 days after injury. An assessment of cerebral autoregulation as part of bedside neuromonitoring in the neurointensive care unit can allow the individualized treatment of secondary injury in a patient with severe TBI. The assessment of cerebral autoregulation is best achieved with dynamic autoregulation methods. Hyperven- tilation, hyperoxia, nitric oxide and its derivates, and erythropoietin are some of the therapies that can be helpful in managing cerebral autoregulation. In this review the authors summarize the most important points related to cerebral pressure autoregulation in TBI as applied in clinical practice, based on the literature as well as their own experience.
    [Show full text]
  • A Theory of Consumer Fraud in Market Economies
    A THEORY OF FRAUD IN MARKET ECONOMIES ADISSERTATIONIN Economics and Social Science Presented to the Faculty of the University of Missouri-Kansas City in partial fulfillment of the requirements for the degree DOCTOR OF PHILOSOPHY by Nicola R. Matthews B.F.A, School of Visual Arts, 2002 M.A., State University College Bu↵alo, 2009 Kansas City, Missouri 2018 ©2018 Nicola R. Matthews All Rights Reserved A THEORY OF FRAUD IN MARKET ECONOMIES Nicola R. Matthews, Candidate for the Doctor of Philosophy Degree University of Missouri-Kansas City, 2018 ABSTRACT Of the many forms of human behavior, perhaps none more than those of force and fraud have caused so much harm in both social and ecological environments. In spite of this, neither shows signs of weakening given their current level of toleration. Nonetheless, what can be said in respect to the use of force in the West is that it has lost most of its competitiveness. While competitive force ruled in the preceding epoch, this category of violence has now been reduced to a relatively negligible degree. On the other hand, the same cannot be said of fraud. In fact, it appears that it has moved in the other direction and become more prevalent. The causes for this movement will be the fundamental question directing the inquiry. In the process, this dissertation will trace historical events and methods of control ranging from the use of direct force, to the use of ceremonies and rituals and to the use of the methods of law. An additional analysis of transformation will also be undertaken with regards to risk-sharing-agrarian-based societies to individual-risk-factory-based ones.
    [Show full text]
  • All-Trans Retinoic Acid Induces Synaptic Plasticity in Human Cortical Neurons
    bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.267104; this version posted September 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. All-Trans Retinoic Acid induces synaptic plasticity in human cortical neurons Maximilian Lenz1, Pia Kruse1, Amelie Eichler1, Julia Muellerleile2, Jakob Straehle3, Peter Jedlicka2,4, Jürgen Beck3,5, Thomas Deller2, Andreas Vlachos1,5,*. 1Department of Neuroanatomy, Institute of Anatomy and Cell Biology, Faculty of Medicine, University of Freiburg, Germany. 2Institute of Clinical Neuroanatomy, Neuroscience Center, Goethe-University Frankfurt, Germany. 3Department of Neurosurgery, Medical Center and Faculty of Medicine, University of Freiburg, Germany. 4ICAR3R - Interdisciplinary Centre for 3Rs in Animal Research, Faculty of Medicine, Justus-Liebig- University, Giessen, Germany. 5Center for Basics in Neuromodulation (NeuroModulBasics), Faculty of Medicine, University of Freiburg, Germany. Abbreviated title: Synaptic plasticity in human cortex *Correspondence to: Andreas Vlachos, M.D. Albertstr. 17 79104 Freiburg, Germany Phone: +49 (0)761 203 5056 Fax: +49 (0)761 203 5054 Email: [email protected] 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.04.267104; this version posted September 4, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ABSTRACT A defining feature of the brain is its ability to adapt structural and functional properties of synaptic contacts in an experience-dependent manner. In the human cortex direct experimental evidence for synaptic plasticity is currently missing.
    [Show full text]
  • SHORT-TERM SYNAPTIC PLASTICITY Robert S. Zucker Wade G. Regehr
    23 Jan 2002 14:1 AR AR148-13.tex AR148-13.SGM LaTeX2e(2001/05/10) P1: GJC 10.1146/annurev.physiol.64.092501.114547 Annu. Rev. Physiol. 2002. 64:355–405 DOI: 10.1146/annurev.physiol.64.092501.114547 Copyright c 2002 by Annual Reviews. All rights reserved SHORT-TERM SYNAPTIC PLASTICITY Robert S. Zucker Department of Molecular and Cell Biology, University of California, Berkeley, California 94720; e-mail: [email protected] Wade G. Regehr Department of Neurobiology, Harvard Medical School, Boston, Massachusetts 02115; e-mail: [email protected] Key Words synapse, facilitation, post-tetanic potentiation, depression, augmentation, calcium ■ Abstract Synaptic transmission is a dynamic process. Postsynaptic responses wax and wane as presynaptic activity evolves. This prominent characteristic of chemi- cal synaptic transmission is a crucial determinant of the response properties of synapses and, in turn, of the stimulus properties selected by neural networks and of the patterns of activity generated by those networks. This review focuses on synaptic changes that re- sult from prior activity in the synapse under study, and is restricted to short-term effects that last for at most a few minutes. Forms of synaptic enhancement, such as facilitation, augmentation, and post-tetanic potentiation, are usually attributed to effects of a resid- 2 ual elevation in presynaptic [Ca +]i, acting on one or more molecular targets that appear to be distinct from the secretory trigger responsible for fast exocytosis and phasic release of transmitter to single action potentials. We discuss the evidence for this hypothesis, and the origins of the different kinetic phases of synaptic enhancement, as well as the interpretation of statistical changes in transmitter release and roles played by other 2 2 factors such as alterations in presynaptic Ca + influx or postsynaptic levels of [Ca +]i.
    [Show full text]
  • Arxiv:1711.09601V4 [Cs.CV] 5 Oct 2018 Unlabeled Data Towards What the Network Needs (Not) to Forget, Which May Vary Depending on Test Conditions
    Memory Aware Synapses: Learning what (not) to forget Rahaf Aljundi1, Francesca Babiloni1, Mohamed Elhoseiny2, Marcus Rohrbach2, and Tinne Tuytelaars1 1 KU Leuven, ESAT-PSI, iMEC, Belgium 2 Facebook AI Research Abstract. Humans can learn in a continuous manner. Old rarely utilized knowl- edge can be overwritten by new incoming information while important, frequently used knowledge is prevented from being erased. In artificial learning systems, lifelong learning so far has focused mainly on accumulating knowledge over tasks and overcoming catastrophic forgetting. In this paper, we argue that, given the limited model capacity and the unlimited new information to be learned, knowl- edge has to be preserved or erased selectively. Inspired by neuroplasticity, we propose a novel approach for lifelong learning, coined Memory Aware Synapses (MAS). It computes the importance of the parameters of a neural network in an unsupervised and online manner. Given a new sample which is fed to the net- work, MAS accumulates an importance measure for each parameter of the net- work, based on how sensitive the predicted output function is to a change in this parameter. When learning a new task, changes to important parameters can then be penalized, effectively preventing important knowledge related to previous tasks from being overwritten. Further, we show an interesting connection between a local version of our method and Hebb’s rule, which is a model for the learning process in the brain. We test our method on a sequence of object recognition tasks and on the challenging problem of learning an embedding for predicting <subject, predicate, object> triplets. We show state-of-the-art performance and, for the first time, the ability to adapt the importance of the parameters based on arXiv:1711.09601v4 [cs.CV] 5 Oct 2018 unlabeled data towards what the network needs (not) to forget, which may vary depending on test conditions.
    [Show full text]
  • Building a “Cross-Roads Discipline” at Mcgill University: a History of Early Experimental Psychology in Postwar Canada
    BUILDING A “CROSS-ROADS DISCIPLINE” AT MCGILL UNIVERSITY: A HISTORY OF EARLY EXPERIMENTAL PSYCHOLOGY IN POSTWAR CANADA ERIC OOSENBRUG A dissertation submitted to the Faculty of Graduate Studies in partial fulfillment of the requirements for the degree of Doctor of Philosophy Graduate Program in Psychology. Graduate Program in Psychology York University Toronto, Ontario October 2020 © Eric Oosenbrug, 2020 Abstract This dissertation presents an account of the development of psychology at McGill University from the late nineteenth century through to the early 1960s. The department of psychology at McGill represents an alternative to the traditional American-centered narrative of the cognitive revolution and later emergence of the neurosciences. In the years following World War II, a series of psychological experiments established McGill as among the foremost departments of psychology in North America. This thesis is an institutional history that reconstructs the origins, evolution, and dramatic rise of McGill as a major center for psychological research. The experiments conducted in the early 1950s, in the areas of sensory restriction, motivation, and pain psychology, were transformative in their scope and reach. Central to this story is Donald O. Hebb, author of The Organization of Behavior (1949), who arrived at McGill in 1947 to find the charred remains of a department. I argue that the kind of psychology Hebb established at McGill was different from most departments in North America; this is developed through a number of interwoven storylines focused on the understanding of a particular character of McGill psychology - a distinctive “psychological style” - and its broader historical importance for Canadian psychology, for North American psychology, and for psychology across the globe.
    [Show full text]
  • Neuromodulated Dopamine Plastic Networks for Heterogeneous Transfer Learning with Hebbian Principle
    S S symmetry Article Neuromodulated Dopamine Plastic Networks for Heterogeneous Transfer Learning with Hebbian Principle Arjun Magotra and Juntae Kim * Department of Computer Science and Engineering, Dongguk University, 30, Pildong-ro 1-gil, Jung-gu, Seoul 04620, Korea; [email protected] * Correspondence: [email protected] Abstract: The plastic modifications in synaptic connectivity is primarily from changes triggered by neuromodulated dopamine signals. These activities are controlled by neuromodulation, which is itself under the control of the brain. The subjective brain’s self-modifying abilities play an essential role in learning and adaptation. The artificial neural networks with neuromodulated plasticity are used to implement transfer learning in the image classification domain. In particular, this has application in image detection, image segmentation, and transfer of learning parameters with significant results. This paper proposes a novel approach to enhance transfer learning accuracy in a heterogeneous source and target, using the neuromodulation of the Hebbian learning principle, called NDHTL (Neuromodulated Dopamine Hebbian Transfer Learning). Neuromodulation of plasticity offers a powerful new technique with applications in training neural networks implementing asymmetric backpropagation using Hebbian principles in transfer learning motivated CNNs (Convolutional neural networks). Biologically motivated concomitant learning, where connected brain cells activate positively, enhances the synaptic connection strength between the network neurons. Using the NDHTL algorithm, the percentage of change of the plasticity between the neurons of the CNN layer is directly managed by the dopamine signal’s value. The discriminative nature of transfer learning Citation: Magotra, A.; Kim, J. fits well with the technique. The learned model’s connection weights must adapt to unseen target Neuromodulated Dopamine Plastic datasets with the least cost and effort in transfer learning.
    [Show full text]